This thread has been locked.

If you have a related question, please click the "Ask a related question" button in the top right corner. The newly created question will be automatically linked to this question.

LM46001-Q1: Analyzing performance of LM46001A-Q1

Part Number: LM46001-Q1
Other Parts Discussed in Thread: LM46001, , TPS7A47

I'm easily confused, so please bear with me. I'm trying to perform an analysis on a circuit that uses a LM46001AQPWPTQ1. I've read the datasheet multiple times and tried to apply the recommended formulas, but I still come away with questions. The designer used a 1nF capacitor on the SS/TRK pin, this appears to fly in the face of what he datasheet states, so, what is the real affect of placing a small cap to ground on this pin?

I also tried to calculate the output ripple and came up with a number ~20 smaller than what WEBENCH calculated (which used a slightly different LM46001 part), FS=500kHz, D=21%, VIN=28, VOUT=6, L=33µH, COUT=58µF.

Using same values, I also calculated a crossover frequency ~1/3 of what WEBENCH calculated.

I need to understand these differences to ensure I'm not going down the primrose path on the analysis, so any enlightenment is much appreciated.

I'm trying to run the analysis for worst case parameters, and, near as I can tell, WEBENCH is limited such that I cannot use it (output capacitance created using 5 caps, won't let me attach CSS, constrains what caps I can use, tries to optimize when I don't want it optimized)

  • Hi WIlliam,

    An EVM userguide will be a good start for you to check the values of your calculated design 

    http://www.ti.com/lit/ug/snvu422/snvu422.pdf

    As for putting a 1nF capacitor on SS/TRK pin, the purpose of this is to control how fast the output voltage is rising. If you put a smaller capacitor value, it will not do anything since it will use internal soft-start value. 

    The inductor calculation will depend on what the VIN being used as well as peak to peak inductor current ripple value desired. In webench normally it calculate inductor based on maximum VIN instead a nominal one, therefore you have to input your parameter accordingly and normally webench uses a ripple value of 30% of Iout value that you entered in.  Therefore in this case datasheet equation might be more tuned to your desired number compared to webench that will take all the operating condition into account. 

    In webench you can click on a component and modify it such that you can select the output capacitor that you use as well as the inductor used. That will add some degree of customization into your design. 

    I hope that helps. Let us know if you have further questions. Normally for each design i would check with EVM of the part itself and see whether my values is within reasonable range

    Thanks

    -Arief

  • Arief, thank you for the reply. Unfortunately the EVM does nothing for me.

    My only goal is to try to analyze performance given the designed values as modified by tolerances, etc. I was trying to use WEBENCH to sanity check the numbers that I was calculating using the formulas in the datasheet. They were not in the same ballpark in some cases which leads to a question of my calculations. I believe I faithfully applied the formulas, so I need to decide whether to believe my numbers.

    If you tell me that I can use the datasheet formulas as provided, then I will accept my calculations.

    While you are correct in that I can click a component and choose another one, WEBENCH limits the choices which, oftentimes, do not match the components actually used. The design is not mine, I'm only paid to try and analyze it. The original designer is not available and, apparently, no preliminary analysis was left behind as a guide to help me understand why certain values were selected. To achieve the extremes of my calculations, I would have to run ~120 designs modifying the components by selecting values close to, but not exactly, what I need. This is not tenable for me.

    I have the TI SPICE transient model for the converter, but it only goes so far in regards to what I'm trying to do and I'm never sure if my results are a product of deficiencies in the model or something I should pay attention to.

    Thank you again, I will keep plugging away.

  • Hello William,

    I understand where you are coming from and we will work together to resolve.

    There is going to be some difference between datasheet and webench.

    Datasheet will contain theoretical calculations, the webench model will be based on experimental data and SPICE engine model.

    What open queries would you like answered? 

  • Well, I think you and Arief answered some of my questions in that I need to take both WEBENCH and the datasheet formulas with a grain of salt.

    I have been given a design and asked to analyze it for performance against worst case conditions. The design uses multiple LM46001-Q1 parts as well as TPS7A4701-Q1 parts. The contract does not expect worst case performance on the power converters and regulators, but I do need to address worst case outputs so that I can apply that against the rest of the circuit. For the circuit in question, the output of the LM46001 feeds a TPS7A47 regulator.

    My thought was to address the design against the theoretical answers from the datasheet formulas which is where I started to get hung up. A few of the questions/concerns:

    1) Using the formulas, I calculated worst case p-p ripple between  1.2 - 1.3mV, given that WEBENCH came back with ~26mV, it causes me to pause. It may very well be the differences in actual capacitors in the design vs what WEBENCH allowed. I'd rather my calculations be correct, but how do I get a warm fuzzy? The design uses five capacitors to form the Cout, so calculations are tedious to get it down to ESR and capacitance as a single set of parameters.

    2) The design used a feed-forward capacitance of 27pF, but the formulas suggested ~47pF. WEBENCH came back with different choices between 27pF and 33pF based on what components it allowed me to pick.

    3) I calculated a crossover frequency between 6 and 12 kHz, zero at 5.7 - 6kHz, and pole at 33 - 35kHz. The datasheet suggested a feed-forward capacitance to get the crossover frequency centered between the zero and pole. WEBENCH had the crossover frequency at around 27kHz. The ranges are sufficiently far that it makes me question my calculations, however, if the calculations are correct is the answer unacceptable from a performance standpoint?

    4) I don't have to address gain and phase margins, because that is inherent in the design, but any advice on how to get a rough magnitude would be helpful.

  • Sorry for delay. I hope all is well William.

    1. I suggest taking experimental data if you want extra level of confidence. EVM is available for order on Ti.com, in addition to output voltage waveforms contained in datasheet.

    2. I expect slight variance between selecting those two componets. For further understanding of CFF calculation: http://www.tij.co.jp/jp/lit/an/snva798/snva798.pdf

    3. Please follow datasheet suggestion

    4. Average SPICE model is available for simulation. In addition, loop response can be validated in Webench. Here is reference for understand loop response in current mode: https://www.ti.com/seclit/ml/slup340/slup340.pdf